Re: Guidelines - events

I think that what realy matters is whether the page is functional 
with and without client side scripts. You can do things that are 
OK with any attribute. 

O.K.: rollovers that change appearance of links when the 
       mouse rolls on them. (as long as the links are 
       noticeable and functional without this effect)

Not OK: a link that leads to an important document but works only 
with scripts on, and gives error message "unknown/unsupporeted URL scheme"
otherwise.

We can't rule against attributes that have no substitute. First the W3C 
should come with platform independent events list, then deprecate 
the current ones.

This is what was done with deprecated markup. HTML4.0 deprecated many
things because there is a better alternative mechanism. Stylesheets, for 
example, or an alternative element (OBJECT instead of APPLET).
Frames that have many problems are not deprecated because some of their
functionality isn't replaced (yet).

Regards,
Nir.

> (Note for UA people - the discussion comes from page authoring and which 
> events should and should not be used - the initial premise was that all 
> device-dependent events should not, although hopefully some could be 
> replaced by 'logical events - ones which can be activated by keyboard, 
> voice input, etc)
> 
> If onKeyPress passes no information about which key was pressed, then it 
> could be mapped, along with onClick, to a new onActivate by UAs, 
> providing a pathway for transition to the 'logical event' which we could ask 
> for in future HTML.
> 
> There may be equivalents to onMouseDown, onMouseUp which should be 
> considered at first in the context of making selections - I don't know 
> how to do this using the keyboard in any browsers although I understand 
> that mapping a mouse to the keyboard is offered by Windows systems. How 
> this would translate to voice systems (or mobile phone systems, with 
> voice and a limited number of keys) I am not sure.
> 
> But I do think that until we have some ide of how to proceed with this 
> the page-authoring guidelines should recommend against it. I know that 
> people will ignore that where they feel they need a rollover highlight 
> for their client, but they can decide that this will not compromise 
> accessibility without throwing out the whole guidelines. (Have some faith 
> in the audience)
> 
> Charles McCathieNevile
> 
> On Wed, 28 Oct 1998, Nir Dagan wrote:
> 
> > I don't think that recommending against using 
> > onClick and similar attributes is that good.
> > The reason being that it may encourage people
> > to do the wrong thing.
> > 
> > For example one can write a gracefully transforming  
> > JavaScript thing with:
> > 
> > <A href="some-url" onClick="some-script">some text</A>
> > 
> > Now if we disallow using onClick he may write 
> > <A href="some-script">some text</A>
> > 
> > that (depending on the script) may work just as before with
> > Javascript on. But doesn't work otherwise.
> > (The second construct is invalid as some-script will not 
> > be a valid URL but validators don't report errors in CDATA 
> > attribue values.)
> > 
> > Another problem with excluding all these atributes, is that they 
> > constitute all the "intristic-event" generic attributes, so it would 
> > be very difficult to use scripts to the extent that script fans 
> > will forget about the WAI.
> > 
> > I think we should find the analogies between mouse events and 
> > keyboard and speach and make a note of that in User-Agent guidelines.
> > 
> > One should remember that the names of the attributes are 
> > platform specific due to the fact that most were introduced as 
> > vendor extentions, but some of them have clear analogies: 
> > onClick and the sugested onActivate, for example.
> > 
> > Regards,
> > 
> > Nir Dagan                            
> > Assistant Professor of Economics      
> > Universidad Pompeu Fabra
> > Barcelona (Spain)
> > 
> > website: http://www.nirdagan.com
> > email: nir.dagan@econ.upf.es
> > 
> > "There is nothing quite so practical as a good theory." 
> > -- A. Einstein
> > 
> >

Received on Wednesday, 28 October 1998 02:39:08 UTC